AI software correctly identified lung cancer with 82% accuracy. Photo: File
London: In an experiment, artificial intelligence (artificial intelligence or AI) has surpassed humans in detecting cancer.
Imperial College London, Cancer Research Center and Royal Marsden NHS Foundation have jointly developed an AI tool. This algorithm can identify lung cancer very quickly and accurately, the results of which were published in the e-Biomedical Journal of The Lancet.
In this context, CT scans of 500 patients were taken and the smallest tumors formed in the lungs were seen from it. The software also saw cancer in places that were invisible to the human eye. It should be noted that the software was previously trained on thousands of CT scans.
In the scans and X-rays, the experts looked for areas called ‘area under the curve’ (AUC) in medical parlance. If the AUC value is one, then the model is excellent, while a reading of 0.5 indicates poor quality.
The AI model detected a nodule, abscess or tumor in the lung with an accuracy of 0.87. Thus, this model can be highly effective in detecting lung cancer. Interestingly, some such details are overlooked by even the most expert doctors.
Overall, 82 percent of the lung cancer suspects the software rang up actually had cancer, as conventional tests later confirmed it.
Experts are optimistic that this way, after the early detection of cancer, it will be possible to help save the lives of thousands of people around the world.
(function(d, s, id){
var js, fjs = d.getElementsByTagName(s)[0];
if (d.getElementById(id)) {return;}
js = d.createElement(s); js.id = id;
js.src = “//connect.facebook.net/en_US/sdk.js#xfbml=1&version=v2.3&appId=770767426360150”;
fjs.parentNode.insertBefore(js, fjs);
}(document, ‘script’, ‘facebook-jssdk’));
(function(d, s, id) {
var js, fjs = d.getElementsByTagName(s)[0];
if (d.getElementById(id)) return;
js = d.createElement(s); js.id = id;
js.src = “//connect.facebook.net/en_GB/sdk.js#xfbml=1&version=v2.7”;
fjs.parentNode.insertBefore(js, fjs);
}(document, ‘script’, ‘facebook-jssdk’));